Johnson & Johnson Innovative Medicine

mRNA language model with a hyperbolic prediction head encoding the codon-amino-acid hierarchy, beating Euclidean baselines on 9 of 10 property tasks.

Multimodal drug-target interaction model aligning SMILES, molecule text, taxonomy, and protein sequence with a Gramian volume contrastive objective.

Gene function prediction over the Gene Ontology graph, inferring new GO annotations for a gene or gene product from the ones it already carries.

Codon-tokenized mRNA language model whose hierarchical loss scores codon errors by synonymity, pretrained on 15.3M curated antibody mRNAs.